how to define map and list in the flatbuffers schema for compile a java class

[paul.ro...@googlemail.com]

hi community :)

i want to use map and list in my next flatbuffers project, but how?

namespace org.example.person;

table Person {
  name:string;
  age:int;
  possession:list;
  children:map<string,int>;
}

root_type Person

sincerly yours
Paul

[paul.ro...@googlemail.com]

i get a email from flatb...@googlegroups.com:
Don't you want children to be a list of Person, e.g.

children:[Person]

Not sure what the int in your map is for, but if it is age, that is already in Person too.

A list is called a "vector" in FlatBuffers, denoted by []. So maybe possessions should be:

possessions:[Possession]
___________________________________________________________________________

thanks for the answer,

how to define a map for dynamic object, also key value pairs?

[Wouter van Oortmerssen]

If the keys in your map are enumerable, then you should really be using a table (with keys as field names), even if few of keys are used.

If the keys are not enumerable, then creating a table that contains both a key and a value, and then storing a vector of those is simplest. If performance is an issue, you could sort this vector by key before you add it to the FlatBuffer, such that you look up elements by binary search when you access the buffer.

You  could even emulate either a binary tree or a hashtable in FlatBuffers, both of which are possible.

I should really add some helper functions for some of these use cases. 

--
You received this message because you are subscribed to the Google Groups "FlatBuffers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flatbuffers...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[戴政]

I use a vector with table, but I can not get the table object by root.Childeren(1);

Could you show me your code?

在 2014年11月25日星期二UTC+8下午10时39分13秒，paul.ro...@googlemail.com写道：

[Vince Harron]

If someone needs to load into a map, (to modify after load, for example) it would be cool if the nodes were sorted in an insertion order that avoided rebalancing.  E.g. breadth first?

[Wouter van Oortmerssen]

reondz: in Java, root.children(1) should return the second table (of type Person) if you use my suggested definition, what exactly is not working?

[Wouter van Oortmerssen]

Vince: that would be best indeed, but tables are stored by the user before the vector/map gets created, so I have no control over that.

An optional conversion to map sounds like a nice feature though.